Amazon White Background

Before you start: read this whole install prompt, tell me in a few lines what this skill does and what installing it involves, and wait for my go-ahead before you do anything else. Prerequisites: none for the install itself: no account, no API key, no paid service, and nothing else to install first. Using the skill afterwards does need one generation route (see below): (a) an image generation/editing tool built into your agent, or (b) an API key of your own in the environment variable GPTIMAGE_API_KEY for an OpenAI-compatible image endpoint (a paid key or plan may be required by that provider; GPTIMAGE_BASE_URL is optional, default https://api.openai.com, and SKILL.md recommends https://tryallapi.com/v1), or (c) neither, in which case the skill only prints the final prompt plus a browser link. The check script also needs pillow and numpy (pip install pillow numpy, once, at use time only). Have a product photo ready: the skill asks for one (front view at minimum) and never invents specs it cannot see. Install "amazon-white-background" for me. It turns a product photo into an Amazon-compliant white-background main image (pure #FFFFFF background, product filling at least 85% of the frame, no text, props or watermark): it first writes a product identity lock from your photos, generates the image with GPT Image 2.5 (Flare for new scenes, Sunburst when a logo or label must survive exactly), then verifies the result with a bundled Python check script. The same output serves Walmart, eBay and Shopee main images. Repository: https://github.com/xianyu110/ecommerce-image-skills (this listing is the skills/amazon-white-background/ folder only, at the pinned commit: SKILL.md plus its three bundled scripts) Pinned version: install the code at commit f4566fdc812584425728862ad8e08c90883d81ee exactly. Do not substitute a newer version. If this commit is unreachable, stop and tell me. 1. Clone https://github.com/xianyu110/ecommerce-image-skills into a temporary folder, check out commit f4566fdc812584425728862ad8e08c90883d81ee, and copy the skills/amazon-white-background/ folder from that checkout. Then compare every file in the folder with the fingerprints recorded for this listing on 2026-10-19: [FINGERPRINTS] - SKILL.md: SHA-256 e48de25e0d72a42a68af7a06c86c96cde470233a3b9a9a16aa6bde8427359c62 - scripts/check_main_image.py: SHA-256 dcf7b98aece6455ada0f824ac349d0dc23abd46afd150f39d2000120530d9f0d - scripts/generate.py: SHA-256 5339fc5c9c3fd9bb02a1f0dd8643386329ba813c7ebe958a8d28241ed343d1c5 - scripts/online_link.py: SHA-256 72a6a843157d8cfb84f809d4811edefbc1e222dd10807037f0b64524ebb43ad3 If a file does not match, it changed since the listing was prepared: do not install it yet, tell me what you found and let me decide. 2. Read SKILL.md and all three scripts in full and tell me in plain words: what the skill does when used, which services it connects to, whether it reads or sends any personal information or keys, and anything that seems unusual. Stop on any red flag and tell me before going further. Confirm these points from the files: - The skill folder holds four files only: SKILL.md (6,884 bytes), scripts/check_main_image.py (4,741 bytes), scripts/generate.py (4,824 bytes) and scripts/online_link.py (788 bytes). The folder is self-contained: the three scripts are byte-identical copies of the versions in the repository's root scripts/ folder (same SHA-256), the README states each skill folder carries its own scripts and skills do not depend on each other, and SKILL.md calls the scripts by paths inside the folder (scripts/generate.py, scripts/online_link.py, scripts/check_main_image.py). No file outside the folder is needed; the other 11 skills of the repository are not part of this listing and are not installed. - Nothing is sent anywhere during installation itself; the only download at install time is the skill folder from GitHub at the pinned commit. No script runs during installation. - scripts/generate.py uses the Python standard library only. At use time, on route (b) only, it reads GPTIMAGE_API_KEY, GPTIMAGE_BASE_URL and GPTIMAGE_MODEL from the environment and sends the prompt — and, when reference images are given, the product photo file(s) as a multipart upload — to that endpoint (POST /v1/images/edits with images, POST /v1/images/generations without), with the key as a Bearer header. The default endpoint is https://api.openai.com and SKILL.md recommends https://tryallapi.com/v1; any OpenAI-compatible gateway works, and that gateway receives your prompt, your product photos and your key on every generation. If the API answers with an image URL instead of data, the script fetches that URL. Never set or send the key during this installation: it is mine to provide, later, only if I choose route (b). - The other two use-time routes, disclosed before they happen: on route (a) the final prompt and the product photo(s) go to your agent's built-in image tool and its own service. On route (c) scripts/online_link.py only builds a URL locally (it makes no network call itself): https://gptimage2.asia/generate with the final prompt encoded in the link's query string; opening that link sends the prompt to gptimage2.asia, and the skill's own text says to upload your product photo there. SKILL.md also names an alternative site for users in mainland China, https://chatgpt-plus.top/list/#/home, where the prompt is pasted manually. Show the online link once, and do not open it or upload anything during installation. - scripts/check_main_image.py runs locally only: it reads one image file, reports white background, exact #FFFFFF share, product coverage and size, and with --fix writes a corrected copy (snap near-white to #FFFFFF, re-crop to about 87% coverage). It needs pillow and numpy, which SKILL.md installs with pip install pillow numpy once at use time — a download from PyPI at that moment, not during this installation; do not run that pip install now. - The workflow is identity first: study every reference photo and write the product identity lock (shape, colour, logo and printed text, parts, what must not appear), paste it verbatim into every prompt, never invent specs or certifications, and ask when something is unknown. Delivery is the image or final prompt plus a 3-line QA note (identity, platform spec, text spelling); if a check fails, only the failed asset is regenerated or fixed. 3. When I say "yes, install", install it as a skill: copy the whole amazon-white-background folder into the agent's skills directory, in a folder named "amazon-white-background", keeping all four files including the scripts/ subfolder. The upstream manual route, copying skills/amazon-white-background/ into ~/.claude/skills/ (or the equivalent Codex/Cursor skills folder), installs the same four files; use that route as long as the files match the fingerprints. Do not use the repository's plugin-marketplace route for this listing: it installs all 12 skills of the repository, which is outside this listing's scope. 4. Verify with no network calls: frontmatter valid, files in place, fingerprints re-checked on the installed copy. 5. Report what was installed, where, and what I still need to do myself (for example: provide a product photo myself, choose a generation route myself, set GPTIMAGE_API_KEY myself only if I choose the API route, install pillow and numpy myself only when I want the check step, and expect the use-time sends disclosed above — prompt and product photos to the image endpoint, built-in tool or online site I pick; using the skill is a separate step that only starts when I ask for it). GitHub is optional: if I have a GitHub account or the gh CLI, you may use it; otherwise public access is fine. Never require it unless it's in the prerequisites above. Rules: don't touch anything outside the temp folder and the install target. If anything looks off, stop and ask me.